No description
Find a file
2025-03-26 19:23:12 +01:00
.github/workflows Add release workflow 2025-03-25 12:32:26 +01:00
doc/images Add GIF demonstration to README and enhance setup instructions 2025-03-25 11:56:50 +01:00
src Refactor fetchCompletion method to improve completion logic and handle empty lines 2025-03-26 19:23:12 +01:00
.editorconfig Initial commit 2025-01-01 21:13:20 +01:00
.eslintignore Initial commit 2025-01-01 21:13:20 +01:00
.eslintrc Initial commit 2025-01-01 21:13:20 +01:00
.gitignore Update .gitignore to exclude main.js file 2025-03-26 10:56:43 +01:00
.npmrc Initial commit 2025-01-01 21:13:20 +01:00
esbuild.config.mjs Rename plugin and update description; adjust esbuild config for Node platform 2025-03-24 22:14:02 +01:00
LICENSE Create LICENSE 2025-03-25 12:24:55 +01:00
manifest.json Bump version to 1.0.1 2025-03-26 11:26:06 +01:00
package-lock.json Update package version to 1.0.1 and remove unused dependencies 2025-03-26 14:16:58 +01:00
package.json Update package version to 1.0.1 and remove unused dependencies 2025-03-26 14:16:58 +01:00
README.md Add GIF demonstration to README and enhance setup instructions 2025-03-25 11:56:50 +01:00
styles.css Refactor SuggestionWidget to support multiple CSS classes and update styles for inscribe-inline-prediction 2025-03-26 11:17:12 +01:00
tsconfig.json Add React support and settings management for Inscribe plugin: update TypeScript configuration, add dependencies, and implement settings page 2025-01-05 21:31:53 +01:00
version-bump.mjs Initial commit 2025-01-01 21:13:20 +01:00
versions.json Bump version to 1.0.1 2025-03-26 11:26:06 +01:00

Obsidian Inscribe Plugin

Inscribe is an AI-powered inline autocompletion plugin for Obsidian. It will generate autocompletions as you type and you can accept or reject them with TAB.

Get started

1. Setup a provider

Inscribe supports following providers:

  • Ollama (local)
  • OpenAI
  • OpenAI compatible

More providers will be added if requested. Or better, feel free to create PR.

2. Customize prompts

You can set both system and user prompts to guide the AI's responses. For user prompt, template variables (mustache) can be used to provide context;

  • {{pre_cursor}}: text before cursor
  • {{post_cursor}}: text after cursor
  • {{active_sentence}}: the current sentence
  • {{last_line}}: last line of the active note

3. Per-Path profile assignments (Optional)

You can create multiple profiles and assign them to specific note paths. For example, you might want a more formal profile for research notes and a more creative one for writing fiction. Corresponding profile will be automatically activated when you open a note matching the assigned path.

Tips

What provider should you use?

Given the high frequency of API calls, using Ollama is highly recommended to reduce costs and improve response times. And for Ollama models, consider using a model bigger than 8B params. For now, gemma3:12b is a good starting point.

Disable completions using path config

Using per-path profile config, you can enable completions only for specific paths. This way, you can avoid annoying interruptions in notes where autocompletion isn't desired.